Abstract

This research work investigates the iron chromatographic displacement from the granular activated carbon (GAC) columns that occurred in the Vars, Ontario groundwater treatment plant. Natural organic matter (NOM) (XAD-8) fractionation showed that 30% of NOM was hydrophobic with the capability of forming complexes with iron. CHELEX-100 isolation indicated that 37–40% of the iron was organically-bound. The quantification of complexed iron performed on GAC microcolumn effluent at iron exhaustion point showed that this iron was neither adsorbed nor split from NOM. This indicates that NOM–iron complexes were not the cause of the observed chromatographic displacement of iron and suggests that competitive adsorption between iron and NOM forced the previously adsorbed iron to be desorbed. Adding greensand treatment prior to the GAC columns resulted in full iron removal, which suggests that NOM–iron complexes must have been either broken by potassium permanganate oxidation and/or adsorbed by the greensand filter.

INTRODUCTION

Some of the most salient characteristics of groundwaters are fairly constant characteristics, high microbiological quality, low or no dissolved oxygen, high mineral content and low natural organic matter (NOM) (MWH 2012). The actual characteristics vary with the soil and rock characteristics, the nature of the aquifer and the aquifer residence time. Accordingly, there is a fairly wide range in the concentrations of the compounds of concern, and in many instances groundwater treatment beyond basic disinfection by chlorination is required. As iron (Fe) and manganese (Mn) are abundant within our planet, many groundwaters have dissolved Fe and Mn concentrations in excess of the aesthetic drinking water guidelines and standards of 0.3 mg Fe/L and 0.05 mg Mn/L, respectively (Health Canada 2017). These standards were established to avoid problems with staining of laundry, staining of bathroom fixtures, and because they may produce undesirable tastes in beverages, such as tea and coffee. Accumulation of Fe and Mn in distribution systems can also cause events in which water has an orange, brown or black colour. Accordingly, many municipal water treatment plants using groundwater incorporate iron and manganese removal processes. If iron alone is a concern and lime softening is utilized, generally there will be sufficient Fe removal without additional treatment (Sommerfeld 1999; MWH 2012). Otherwise, the treatment may include aeration and/or pre-oxidation (to oxidize Fe and Mn to more insoluble forms) followed by separation by filtration through sand, anthracite or manganese-coated media. Aeration is not effective for oxidizing Mn or NOM complexed Fe, so chemical oxidation is preferred. As Mn is more difficult to oxidize than iron, potassium permanganate is the preferred oxidant when Mn is present (Logsdon et al. 1999).

Although much less common than Fe and Mn, some groundwaters can have significant levels of NOM (total organic carbon (TOC) (>2 mg/L). These include surface aquifers with swampy recharge zones, such as the Biscayne Aquifer beneath Southeastern Florida which is recharged by the Everglades. High NOM concentrations are a concern because their chlorination can result in the formation of disinfection byproducts (DBPs) at concentrations above the health-related standards (Health Canada 2017). The most economical method of removing the NOM is coagulation/flocculation/sedimentation using a coagulant and a coagulant dose for optimized NOM removal. This approach leads to the creation of sludges that require significant additional operations and can be a maintenance problem for small communities without full-time operators. Although it will likely result in a higher cost, adsorption of the NOM in a granular activated carbon (GAC) can remove the NOM and will create fewer operations and maintenance concerns.

In 1996, the City of Ottawa commissioned a new well field and water treatment system in the town of Vars, Ontario (population of 700) to replace the numerous local private wells that had bacterial contamination problems (City of Ottawa 1996). The well water had a substantial NOM concentration (4–6 mg TOC/L) and colour (≈20 colour units). Based on these two parameters, the design engineers decided to incorporate GAC filters to treat the groundwater in Vars. Although the well water had significant concentrations of iron (≈2 mg/L), the final water treatment system design did not incorporate an iron removal strategy because it was considered to be an aesthetic issue rather than a health concern. The treatment system consisted of two 22-m deep production wells, two particle cloth pre-filters, two granular activated carbon (GAC) columns, post-chlorination and a water reservoir to store the treated water before delivery to the distribution system. The GAC system was designed with an empty bed contact time (EBCT) of 33 minutes per column at an average design flow rate of 10.6 L/s.

The treatment system started production on January 1st, 1996. The GAC columns were initially operated in series where column ‘B’ was fed by the effluent from column ‘A’. The removal of TOC by the GAC columns was nearly complete and gradually decreased with time. After 204 days of operation and 1,620 bed volumes of water treated, the effluent concentration of iron from the lead GAC column increased to levels greater than those of the feed. This phenomenon is called the chromatographic effect. Given that the GAC was intended for TOC removal and the effluent of the second column still contained no iron and very little NOM, the GAC column in the series system was kept in operation. After 443 days and 3,520 bed volumes of water treated, it was decided to bypass column A due to the chromatographic displacement of iron, and operate only column B that was still achieving full iron removal. After 507 days (i.e., 4,030 bed volumes treated), the iron concentration from column B started to increase until the iron concentration of the effluent exceeded that of the influent. After GAC treatment, the iron in water was oxidized through chlorination and the exposure to air in the treatment plant's reservoir. Consequently, oxidized iron precipitated onto the reservoir floor and on the inner surface of the pipes connecting the treatment facility to the consumers. Eventually, some of Vars' residents had discoloured water (orange and brown) coming out of their faucets as the iron precipitate was transported out to the distribution system. Accordingly, the city had to perform a number of distribution system flushings and line pigging to solve the coloured water problems.

Over the years (1996–2002), the activated carbon was replaced several times with different brands of activated carbon (NORIT, PICA and Filtrasorb F-400) for both columns A and B. All the types of activated carbons used over the years behaved similarly in terms of iron removal. As an example, Figure 1 shows the performance of the Vars GAC column for a typical run. It presents the treatment cycle with Filtrasorb F-400 activated carbon in which the iron breakthrough started after 42 days, the effluent iron concentration exceeded that of the raw water after approximately 70 days, and thereafter the GAC column eluted rather than removed the iron.

Figure 1

Performance of Vars GAC column using Filtrasorb F-400 for iron removal (June 2001–May 2002).

Figure 1

Performance of Vars GAC column using Filtrasorb F-400 for iron removal (June 2001–May 2002).

Table 1 summarizes the performance of each carbon type used in Vars in the years 1996, 1998 and 2001. In 1996 when NORIT carbon was used, both GAC columns (A and B) were run in series, whereas only one GAC column was run at a time when PICA and F-400 GACs were utilized. Although F-400 GAC started the iron breakthrough after 42 days, which was earlier than the other two GAC brands (204 and 176 days for NORIT and PICA, respectively), the bed volumes (BV) of water treated at the start of the iron breakthrough were of the same order of magnitude (1,620, 1,395 and 1,026 for NORIT, PICA and F-400, respectively). This similarity is due to the fact that the flow rate of the groundwater to be treated increased when the F-400 GAC was used. Also, PICA and NORIT carbons were exhausted to a greater extent in terms of NOM (C/Co > 0.5) before the start of the iron breakthrough. Table 1 shows that the values of C/Co of NOM at the start of iron breakthrough were 0.67 and 0.8 using PICA and NORIT GAC, respectively. However, F-400 GAC was capable of removing more NOM even after reaching iron breakthrough (C/Co (TOC) ≈ 0.33). The higher value of bed volumes of water treated by F-400 GAC (BV = 4,253) at C/Co (TOC) = 0.5 compared to the other GAC brands (BV = 1,120 and 840 for NORIT and PICA, respectively) indicates that F-400 carbon was superior in terms of removal of NOM even in the presence of iron in the water to be treated. However, F-400 was not capable of solving the iron problem.

Table 1

Performance of column A using different activated carbon types at Vars treatment plant

  NORIT PICA Filtrasorb (F-400) 
Start of the run 01/01/1996 24/02/1998 11/06/2001 
End of the run 19/03/1997 09/11/1998 27/05/2002 
Bed volume of water treated at C/Co (TOC) = 0.5 1,120 840 4,253 
Start of Fe breakthrough (days) 204 176 42 
Bed volumes of water treated at the start of iron breakthrough 1,620 1,395 1,026 
(C/Co) of TOC concentration at the start of iron breakthrough 0.67 0.80 0.33 
  NORIT PICA Filtrasorb (F-400) 
Start of the run 01/01/1996 24/02/1998 11/06/2001 
End of the run 19/03/1997 09/11/1998 27/05/2002 
Bed volume of water treated at C/Co (TOC) = 0.5 1,120 840 4,253 
Start of Fe breakthrough (days) 204 176 42 
Bed volumes of water treated at the start of iron breakthrough 1,620 1,395 1,026 
(C/Co) of TOC concentration at the start of iron breakthrough 0.67 0.80 0.33 

The GAC replacements for such a small size treatment plant made the operating cost per cubic metres of Vars systems much higher than other water treatment systems operated by the City of Ottawa. Accordingly, in June 2002, the City of Ottawa modified the treatment system by the addition of a potassium permanganate (KMnO4) oxidation system and greensand filtration prior to the GAC adsorption columns. The oxidation system consisted of a pump which introduced KMnO4 solution into a mixing tank to oxidize the iron. The greensand filter removed the iron from the water by trapping the ferric hydroxide flocs formed and by adsorbing iron onto the surface of greensand that is coated with manganese oxide. This process significantly improved the performance of the GAC columns. For runs using F-400, the bed volume of water treated to reach 50% TOC removal increased from 5,060 to 11,900.

The objective of this research is to gain a better understanding of both NOM and iron removals at Vars treatment facility. This type of groundwater is complex as it contains free ferrous iron (Fe(II)), ferric iron (Fe(III)), NOM-complexed iron and NOM. The chromatographic displacement of iron from the GAC columns may be associated with the dissociation and leaching of NOM–iron complexes, or possibly with competitive adsorption. The specific objectives of this research are to gain a better understanding of the GAC treatment system by: (1) characterizing the quality of Vars groundwater; (2) quantifying the role of NOM-complexed iron fraction and other fractions on the operation of the adsorption system; and (3) characterizing the greensand-treated water and the effect of this treatment process on the NOM present in water. A sister paper will study the possibility that competitive adsorption of NOM and iron causes the chromatographic effect.

MATERIALS AND METHODS

Groundwater quality characterization

Vars groundwater quality was characterized before and after treatment. Water characterization was performed on water collected from different locations along the treatment line: untreated well water, prior to entering the greensand filter (but subjected to potassium permanganate oxidation) and immediately after greensand filtration. Water samples collected from these three sampling locations were characterized using step (I) and only the untreated well water was characterized using step (II), as shown in Figure 2.

Figure 2

Water characterization analysis sequence.

Figure 2

Water characterization analysis sequence.

Organic free water processed by an ultrapure water system was used for the preparation of the standard solutions as well as for glassware cleaning. The natural organic matter was measured in terms of TOC concentrations, which were determined using a UV-persulfate oxidation-based TOC analyser. In addition, UV absorbance at 254nm was determined using a UV spectrophotometer. A 100-mm path-length quartz cell was used instead of a 10-mm cell in order to accurately measure the UV254 absorbance of the samples with low TOC concentrations. The specific UV-absorption (SUVA) of Vars groundwater was calculated by dividing the UV254 absorbance of the sample in (1/m) by the TOC of the sample in (mg/L). Ferric and ferrous iron concentrations were measured via a two-step procedure. An axially inductively coupled plasma atomic absorption spectrometer (ICP-OES) was used to determine the total iron concentration present in the water samples. The ferrous iron concentrations were measured at the site in order to avoid any type of oxidation that may occur during sample transportation. These measurements were performed by the phenanthroline method using a portable spectrophotometer. The ferric iron concentration was calculated by subtracting the ferrous concentration from the total iron concentration. The pH and Eh values were measured using a pH meter. The measurements of colour of the water samples were performed via a spectrophotometer using the built-in colour scale. Analysis of turbidity, alkalinity and hardness were performed in accordance with Standard Methods(APHA 1998).

Organic bound iron analysis

CHELEX-100TM is a non-NOM-adsorbing cation-exchange resin used to separate the free iron from the organically-bound iron present in Vars groundwater. Researchers have found that CHELEX-100 was a strong complexing resin and it was preferred for metal removal from water (Pesavento et al. 2001; Soylak 2004; Alberti et al. 2005; Biesuz et al. 2006). Moreover, other researchers have confirmed the high preference of CHELEX-100 resin for iron (Jensen et al. 1998; Patrick & Verloo 1998). In a study by Schmitt et al. (2003) on metal–NOM complexes, it was found that iron was strongly bound to NOM. That bond was found to be stable at a pH value higher than 6.8, and therefore the CHELEX-100 resin was not capable of dissociating it. The water sample extracted from Vars groundwater was filtered through a column packed with 150 mg of CHELEX-100 resin using a peristaltic pump at a flow rate of 5 mL/min (i.e., a hydraulic loading rate of 3.2 m/hr). Glass wool was placed at the bottom of the column to prevent the loss of resin. The free iron in the water samples was adsorbed onto the CHELEX-100 resin, whereas the organically-bound iron flowed through the CHELEX-100 column. The tests were conducted until the breakthrough was reached and then the resin was replaced before starting the next run.

XAD-8 fractionation

Fractionation of the natural organic matter into hydrophobic and hydrophilic fractions was performed using the methyl methacrylate resin XAD-8™ following the procedure outlined by Thurman & Malcolm (1981). This test quantifies the hydrophobic fractions of NOM in water samples. These fractions are known to have the capability of forming soluble complexes with inorganic species such as iron. A 200 mL of Vars groundwater sample had its pH adjusted to 2 (±0.1) using 10% 0.1 M HCl and was then passed three times through a column filled with XAD-8 resin. The hydrophilic fraction passed through the column, while the hydrophobic fraction was adsorbed onto the resin. The hydrophobic fraction was later eluted off the column with a 0.1 N NaOH solution. After analyzing the samples using the TOC analyser, a mass balance was conducted to calculate the fractions in the groundwater.

Ultrafiltration fractionation (UF)

NOM was fractionated using ultrafiltration membranes in a batch pressure-driven dead-end filtration system. In this method, the sample was sequentially filtered using a 400 mL Amicon stirred ultrafiltration cell with different nominal molecular weight cut-offs (NMWCO) ultrafiltration membranes. These UF membranes had NMWCO of 500, 1,000, 3,000, 5,000, 10,000 and 30,000 Da. The water was driven through the membranes by high purity nitrogen gas in order to limit the oxidation of iron present in the samples. The fractionation was performed by the in-series ultrafiltration fractionation method, which has been extensively used in the literature (Kilduff et al. 1996; Pelekani & Snoeyink 1999; Tadanier et al. 2000; Kitis et al. 2002). The mass fractions of NOM and iron of a particular molecular weight were obtained by subtracting the TOC and total iron concentration values of the filtrate of one membrane from the filtrate of the next, larger NMWCO membrane. UF fractionation was used to characterize the raw water as well as the CHELEX-100 treated water (i.e., the organically-bound iron in the raw water).

Microcolumn experiment

Since the current Vars treatment system no longer uses GAC filters to treat the raw well water directly, it was not possible to assess the impact of NOM-complexed iron on the breakthroughs with chromatographic overshoots. Therefore, microcolumn experiments were used to assess the removal of complexed iron at different points within the iron breakthrough. This experiment would identify whether the GAC has the capability to dissociate the bond between the NOM and iron during the adsorption process.

The column experiments shown in Figure 3 consisted of two peristaltic pumps, two glass microcolumns (height of 40 cm and diameter of 1.1 cm). Small particle sized GAC was used to accelerate the kinetics. The design parameters of the GAC microcolumn (flow rate, GAC particle size and depth) satisfied the criteria to decrease the iron breakthrough time and to avoid any wall effect (column diameter to GAC particle median diameter ratio equal to 69). The GAC column was packed with 7 g of 80 × 100 mesh size Calgon F-400, a bituminous coal-based activated carbon used at Vars. The carbon depth inside the column was 13 cm, providing an EBCT of approximately 1.2 minutes and a hydraulic loading rate of 6.3 m/hr (corresponding to a flow rate of 10 mL/min). To minimize iron oxidation in these experiments and maintain anaerobic raw water conditions, the microcolumn system was operated using water from one of Vars' raw groundwater pumps. The small EBCT of the microcolumn (EBCT = 1.2 min) was chosen to limit the length of the run to approximately 1 week. However, the magnitude of the EBCT was not critical since the objective of the tests was to sample the NOM–iron complexes removal at different points along the iron breakthrough and not to use it for scale-up. The effluent from the GAC was collected regularly and its iron (both ferrous and total iron) concentrations were measured. Upon reaching one of the chosen points on the iron breakthrough curve, the NOM–iron complexation in the GAC effluent was quantified by passing the GAC column effluent through a CHELEX-100 column. The CHELEX-100 testing procedure was that described earlier in this section and the effluent was tested for TOC, ferrous and total iron.

Figure 3

Setup of the microcolumn experiment.

Figure 3

Setup of the microcolumn experiment.

RESULTS AND DISCUSSION

Water quality characteristics

The quality of Vars groundwater was characterized during two visits in the winter of 2008 (January 10th and February 14th, 2008) and a third visit in the spring of the same year (April 25th, 2008). Table 2 summarizes the characteristics on one of the three visits (February 14th, 2008); the results obtained from all the visits were very similar. The iron concentrations from the two wells were slightly different, which was somewhat surprising given that the two wells have the same depth and are about 30 metres apart. However, the TOC concentration of Vars groundwater was approximately 3.9 mg/L, which is relatively high for a groundwater. This is likely due to the adjacent wetlands. According to the results of the temperature, pH and redox potential values obtained from the preliminary analysis of Vars groundwater and the values on the potential-pH diagram of iron (known as Pourbaix diagram) presented by Brookins (1988), the iron present in this groundwater falls around the boundary line between the soluble divalent ferrous iron Fe(II) and the precipitate ferric hydroxide Fe(OH)3(s). However, the analysis of ferrous iron in the groundwater from the three preliminary visits concluded that most of the iron in Vars groundwater was in the form of Fe(II). Vars groundwater has a relatively steady temperature of 8–9 °C. The water from both wells had an average value of 13 colour units with the values remaining unchanged over a span of the three visits to Vars. The SUVA values found from the three visits were approximately 2 L/mg.m for both wells 1 and 2. According to Edzwald & Tobiason (1999), for raw waters with SUVA below 2 L/mg.m, NOM is normally dominated by mostly low molecular weight and substances with low hydrophobicity. Such a hypothesis would be tested later using XAD-8 and ultrafiltration fractionation.

Table 2

Characteristics of Vars groundwater (February 14th, 2008)

Parameters Units Well #1 Well #2 
pH  7.5 7.6 
Eh (Volts) 0.21 0.21 
Pe  3.6 3.6 
Temperature (°C) 8.5 8.7 
Turbidity (N.T.U) 0.22 0.23 
Total alkalinity mg/L (as CaCO3218 196 
Total hardness mg/L (as CaCO3231 204 
Total Fe (mg/L) 1.07 1.37 
Fe(II) (mg/L) 1.04 1.36 
Fe(III) (mg/L) 0.03 0.01 
TOC (mg C/L) 3.96 3.87 
SUVA (L/mg.m) 2.01 2.02 
Parameters Units Well #1 Well #2 
pH  7.5 7.6 
Eh (Volts) 0.21 0.21 
Pe  3.6 3.6 
Temperature (°C) 8.5 8.7 
Turbidity (N.T.U) 0.22 0.23 
Total alkalinity mg/L (as CaCO3218 196 
Total hardness mg/L (as CaCO3231 204 
Total Fe (mg/L) 1.07 1.37 
Fe(II) (mg/L) 1.04 1.36 
Fe(III) (mg/L) 0.03 0.01 
TOC (mg C/L) 3.96 3.87 
SUVA (L/mg.m) 2.01 2.02 

Identification of organically-bound iron using CHELEX-100

As shown in Figure 4, the CHELEX-100 fractionations showed that the percentage of the organically-bound iron in Vars raw water was 37%. The same experiment was repeated in the spring of 2008 (April 25th, 2008) and the percentage of the organically-bound iron was found to be 40% for well #2. The TOC of the CHELEX-100 column effluent was measured throughout the experiments. As expected, there were no changes in the concentrations of the groundwater TOC levels after passing through the column, indicating that none of the organics were removed by the CHELEX-100 resin. For the groundwater samples, the percentage of free iron removal started to decrease after 50 minutes, indicating the decreasing ability of the resins to hold more of the free iron. Based on these results, only samples of the water passing through the columns that were collected during the first 30 minutes of the experiment were used in the ultrafiltration fractionation in order to identify which NOM fractions were bound to the iron.

Figure 4

Iron reduced concentration of Vars groundwater passing through CHELEX-100 column (mass of CHELEX-100 used = 150mg, flow rate = 5mL/min).

Figure 4

Iron reduced concentration of Vars groundwater passing through CHELEX-100 column (mass of CHELEX-100 used = 150mg, flow rate = 5mL/min).

The organically-bound iron found in the groundwater collected from the CHELEX-100 column effluent was fractionated using the ultrafiltration fractionation method in order to identify which NOM fractions were bound to the iron. Figure 5 shows the results of the UF fractionation of CHELEX-100 filtered Vars groundwater which contains both Fe–NOM complexes as well as uncomplexed NOM. Samples collected from each molecular weight cut-off ultrafiltration were analysed for TOC and total iron. The results presented in Figure 5 illustrate that most of the (organically-bound) iron was in NOM fractions that are larger than 5 kDal, indicating that only large molecular weight fractions of NOM were capable of forming complexes with the iron present in Vars groundwater. This coincides with the findings of Gu et al. (1995), who found that the NOM bound with iron was predominantly a high molecular weight fraction that is hydrophobic in nature.

Figure 5

UF fractionation of the complexed iron in Vars groundwater passing through the CHELEX-100 column.

Figure 5

UF fractionation of the complexed iron in Vars groundwater passing through the CHELEX-100 column.

Identification of complexed organics through XAD-8 isolation

The XAD-8 fractionation was conducted on water samples collected on three different sampling days. The hydrophilic NOM in the Vars groundwater was found to be 75%, 73% and 78% for the samples collected in January, February and April 2008, respectively. This signifies that less than 30% of the NOM was hydrophobic (including humic and fulvic acids). Such a finding confirms the SUVA results of the preliminary analysis of Vars groundwater, which revealed that NOM is dominated by mostly low molecular weight and substances with low hydrophobicity. This is consistent with the findings of Thurman (1985), who found that different deep groundwaters contained 12–33% hydrophobic (humic and fulvic) acids. He hypothesized that the long residence time of natural organic matter in groundwater causes the hydrophobic fractions to be either adsorbed onto the aquifer solids or degraded by bacteria into simpler organic acids. The hydrophobic fractions were found to have the capability of forming soluble complexes with inorganic species, such as iron, through the phenolic and carboxylic functional groups (Gjessing 1976; Stevenson 1982; Perdue 1985; Christman et al. 1989). Accordingly, the maximum amount of NOM that could be complexed with iron is 30%, which represents the hydrophobic fraction of Vars groundwater.

Microcolumn experiment

Microcolumn experiments were conducted to investigate the potential relationship between NOM–iron complexes and GAC column performance, particularly the chromatographic displacement of iron. Figure 6 shows the breakthrough of both NOM and iron represented by the left and right y-axes, respectively. The TOC of the influent ranged between 4.23 and 4.74 mg/L and the iron ranged from 0.95 to 1.10 mg/L. The column experiment lasted for 8 days. At the end of the breakthrough, the effluent iron reduced concentration was 0.92 whereas the effluent TOC reduced concentration was 0.6.

Figure 6

TOC and iron effluent concentrations for the Vars microcolumn test.

Figure 6

TOC and iron effluent concentrations for the Vars microcolumn test.

The quantification of the NOM–iron complex removals was carried out on days 2, 5 and 7 on the breakthrough curve shown in Figure 6. These times represent sampling before the iron breakthrough (day 2), early in the iron breakthrough (day 5) and near exhaustion (day 7). The dashed line in the figure represents the average value between 11 and 15% of complexed iron in the raw water based on multiple samplings. The partial removal of the complexed-iron of the first two samplings (before the breakthrough and early in the breakthrough) indicated that complexed iron was either adsorbed into the activated carbon or split from NOM, and both components were adsorbed separately. The results from the last sampling on the 7th day show that complexed iron was neither adsorbed nor split from the NOM since the concentrations of complexed iron in the raw water and the GAC effluent were almost the same. Such a finding indicates that chromatographic displacement of iron from the GAC observed at Vars was not related to NOM–iron complexes. Another potential reason for the chromatographic displacement of iron could be the competitive adsorption between iron and NOM that forces the iron to be desorbed from the GAC adsorbing sites.

The microcolumn was operated for a few additional days after reaching the iron's point of exhaustion (after the 8th day) but the chromatographic displacement of iron was not observed. Instead, there was a clear visible accumulation of iron within the top section of the microcolumn, which impacted its operation and the test had to be terminated. Given the smaller GAC granule sizes (80 × 100 mesh size), this phenomenon of iron accumulation is more significant in a microcolumn test than in full-scale facilities. This also indicates that the microcolumn tests are not a good tool for simulating adsorption of iron-laden waters.

The difference between the 11 to 15% iron complexed with NOM to approximately 30% measured during the 2008 characterization stage was due to the fact that the microcolumn experiment was performed 3 years later (September 2011). This was, nonetheless, rather surprising given that in these 3 years the iron concentrations, the NOM concentrations and the hydrophilic fraction of the NOM remained fairly constant.

Impact of greensand treatment

As expected, the greensand pretreatment (permanganate oxidation) was successful. The dark bars in Figure 7 show that the permanganate treatment was effective in oxidizing the ferrous iron in the well groundwater into ferric iron. The clear bars show that the greensand filter removed the oxidized iron to levels below the 0.3 mg/L drinking water standard (WHO 2004).

Figure 7

Iron concentrations at different locations along the greensand pretreatment (February 14th, 2008).

Figure 7

Iron concentrations at different locations along the greensand pretreatment (February 14th, 2008).

The TOC concentrations indicate that the greensand treatment removed only 9 to 13% of the NOM (Table 3). This is consistent with the work of Li et al. (2003), who found negligible TOC removals by greensand treatment of a central Illinois groundwater (TOC = 2.17 mg/L). They hypothesized that the potassium permanganate oxidation reaction of the organically-bound iron would form iron–NOM particulate matter that would be removed through greensand filtration.

Table 3

TOC removal due to greensand pretreatment

  TOC removal (%) greensand #1 TOC removal (%) greensand #2 
1st Visit (January 10th, 2008) 11.93 10.61 
2nd Visit (February 14th, 2008) 10.33 11.79 
3rd Visit (April 25th, 2008) 12.85 9.05 
  TOC removal (%) greensand #1 TOC removal (%) greensand #2 
1st Visit (January 10th, 2008) 11.93 10.61 
2nd Visit (February 14th, 2008) 10.33 11.79 
3rd Visit (April 25th, 2008) 12.85 9.05 

The results of the UF fractionation of Vars groundwater and the greensand treated water (Figure 8) indicate that most of the NOM fractions that were removed by the greensand pretreatment were with a molecular weight of less than 1 kDal. Another effect of the KMNO4 treatment on the NOM was the agglomeration of certain fractions (between 1 and 5 kDal) after the removal of most of the iron present in Vars groundwater. Since the greensand filter removed all of the iron, the NOM–iron complexes must have been either broken by the potassium permanganate oxidation or were adsorbed by the greensand filter. Unfortunately, due to the high permanganate concentrations in the greensand column influent, it was impossible to determine the NOM-complexed iron at this point.

Figure 8

UF fractionation of NOM in Vars groundwater and the greensand-treated water (February 14th, 2008).

Figure 8

UF fractionation of NOM in Vars groundwater and the greensand-treated water (February 14th, 2008).

CONCLUSIONS

This study investigated the role of organically-bound iron on the iron breakthrough experienced in the Vars, ON, GAC columns. During the initial sampling campaign, between 37 and 40% of the iron in Vars groundwaters was organically-bound and the rest was in the form of free iron. It was also found that 30% of the NOM in Vars groundwater are hydrophobic with the capability of forming complexes with the iron present in the groundwater.

The GAC microcolumn experiments demonstrated that that NOM–iron complexes were not the cause of the chromatographic displacement of iron because there was no change in the complexed iron concentrations during the near exhaustion sampling. Also, it shows that GAC adsorption did not have the capability of dissociating the NOM–iron complexes. Based on these findings, competitive adsorption between NOM and iron seems the more likely cause of the observed chromatographic effect.

As expected, the added greensand pretreatment resulted in the removal of iron, and it also removed 9–13% of the NOM. The full removal of iron during the greensand treatment indicates that either NOM–iron complexes were broken by the potassium permanganate oxidation and/or were adsorbed by the greensand filter.

ACKNOWLEDGEMENT

The authors gratefully acknowledge King Fahd University of Petroleum and Minerals (KFUPM) and the Natural Science and Engineering Research Council Canada (NSERC) for their financial support, and the City of Ottawa for giving them access to Vars Treatment Plant. The cooperation of Penny Wilson and Mark Silas of the City of Ottawa is greatly appreciated. Also, the authors would like to thank Lara Gallo for her help in performing some of the experiments in this study.

REFERENCES

REFERENCES
Alberti
G.
,
D'Agostino
G.
,
Palazzo
G.
,
Biesuz
R.
&
Pesavento
M.
2005
Aluminium speciation in natural water by sorption on a complexing resin
.
Journal of Inorganic Biochemistry
99
(
9
),
1779
1787
.
APHA, AWWA & WEF
1998
Standard Methods for the Examination of Water and Wastewater
,
20th edn
.
American Public Health Association, American Water Works Association, Water Environment Federation
,
Washington, DC
.
Biesuz
R.
,
Alberti
G.
,
D'Agostino
G.
,
Magi
E.
&
Pesavento
M.
2006
Determination of cadmium(II), copper(II), manganese(II) and nickel(II) species in Antarctic seawater with complexing resins
.
Marine Chemistry
101
(
3–4
),
180
189
.
Brookins
D. G.
1988
Eh-pH Diagrams for Geochemistry
.
Springer-Verlag
,
Berlin
.
Christman
R. F.
,
Norwood
D. L.
,
Seo
Y.
&
Frimmel
F. H.
1989
Oxidative degradation of humic substances from freshwater environments
. In:
Humic Subtances II
(
Bermingham Hayes
M. H.
, ed.).
John Wiley & Sons
,
New York
.
City of Ottawa
1996
Annual Report for Vars Communal Well System
.
Water Quality Division
, City of Ottawa,
Ottawa, ON
.
Edzwald
J. K.
&
Tobiason
J. E.
1999
Enhanced coagulation: US requirements and a broader view
.
Water Science & Technology
40
(
9
),
63
70
.
Gjessing
E. T.
1976
Physical and Chemical Characteristics of Aquatic Humus
.
Ann Arbor Science Publishers
,
Ann Arbor, MI
.
Gu
B. H.
,
Schmitt
J.
,
Chen
Z.
,
Liang
L. Y.
&
McCarthy
J. F.
1995
Adsorption and desorption of different organic-matter fractions on iron-oxide
.
Geochimica et Cosmochimica Acta
59
(
2
),
219
229
.
Health Canada
2017
Guidelines for Canadian Drinking Water Quality-Summary Table
.
Water and Air Quality Bureau, Healthy Environments and Consumer Safety Branch, Health Canada
,
Ottawa, Ontario
.
Jensen
D. L.
,
Boddum
J. K.
,
Redemann
S.
&
Christensen
T. H.
1998
Speciation of dissolved iron(II) and manganese(II) in a groundwater pollution plume
.
Environmental Science & Technology
32
(
8
),
2657
2664
.
Li
Q. L.
,
Snoeyink
V. L.
,
Mariñas
B. J.
&
Campos
C.
2003
Elucidating competitive adsorption mechanisms of atrazine and NOM using model compounds
.
Water Research
37
(
4
),
773
784
.
Logsdon
G.
,
Hess
A.
&
Horsley
M.
1999
Guide to selection of water treatment processes
. In:
AWWA's Water Quality & Treatment
,
5th edn.
(
Lettermann
R. D.
, ed.).
McGraw-Hill
,
New York
.
MWH
2012
Water Treatment: Principles and Design
,
3rd edn
.
John Wiley & Sons
,
New York
.
Pelekani
C.
&
Snoeyink
V. L.
1999
Competitive adsorption in natural water: role of activated carbon pore size
.
Water Research
33
(
5
),
1209
1219
.
Perdue
E. M.
1985
Acidic functional groups of humic substances
. In:
Humic Substances in Soil, Sediments, and Water Geochemistry, Isolation, and Characterization
(
Aiken
G. R.
, ed.).
John Wiley & Sons
,
New York
.
Sommerfeld
E. O.
1999
Iron and Manganese Removal Handbook
.
American Water Works Association
,
Denver, CO
.
Stevenson
F. J.
1982
Methods of Soil Analysis: Nitrogen – Organic Forms
.
American Society of Agronomy and Soil Science Society of America
,
Madison, WI
.
Thurman
E. M.
1985
Organic Geochemistry of Natural Waters
.
Kluwer Academic
,
Boston, MA
.
Thurman
E. M.
&
Malcolm
R. L.
1981
Preparative isolation of aquatic humic substances
.
Environmental Science & Technology
15
(
4
),
463
466
.
WHO (World Health Organization)
2004
Guidelines for Drinking Water Quality, 3rd edn, Vol. 1, Recommendations
.
WHO
,
Geneva
,
Switzerland
.